Rag And Retrieval medium
Your retrieval stage has an 800 ms p95 budget. The team reranks the top 1,000 first-stage candidates with a cross-encoder, which adds about 2 seconds. How do you meet the budget while keeping most of the quality gain?
- A Drop the reranker entirely and pass the top 1,000 first-stage candidates straight into the generator's context.
- B Rerank only the top 50-100 first-stage candidates, since cross-encoder cost scales linearly with candidate count.
- C Replace the cross-encoder with a second bi-encoder pass, since rescoring with cached vectors gives the same quality.
- D Precompute cross-encoder scores for every chunk at ingestion time so that query-time reranking becomes a lookup.
Answer: B. Rerank only the top 50-100 first-stage candidates, since cross-encoder cost scales linearly with candidate count.
A cross-encoder scores each (query, chunk) pair at query time, so its cost grows with the number of candidates. Reranking 50-100 candidates typically adds tens to a few hundred milliseconds. Precomputing scores is impossible because they depend on the query. The tradeoff of a smaller pool is that gold chunks ranked below the cutoff are lost, so confirm recall@100 is high before shrinking it.
One system design question a day, with spaced repetition so the ones you miss come back.
Get the appRelated questions
- You must index 100M chunks using a 3,072-dimension embedding in fp32, served from RAM via HNSW. The raw vectors alone take about 1.2 TB before graph overhead. You need to cut memory several-fold while losing at most a point or two of recall@10. Which approach fits?
- A multi-tenant SaaS app stores all tenants in one vector index with a tenant_id field. Queries fetch the top 20 chunks by similarity and then drop chunks from other tenants. Small tenants often get 0-2 results. Why does this happen, and what is the fix?
- Runbooks are split into fixed 300-token chunks with no overlap. When a procedure step starts near the end of one chunk and finishes in the next, answers come back incomplete. What is the lowest-cost fix?
- Your offline eval shows recall@5 of 92% on a labeled set, so the gold passage is in the context for nearly every question. Yet human graders mark 35% of answers wrong, and the wrong answers often ignore or contradict the gold passage. Where should the team invest next?
- Your product requires every RAG answer to include clickable source citations that users can verify. How should those citations be produced?
- A runbook RAG system uses fixed 500-token chunks with no overlap. Users who ask for a multi-step recovery procedure get answers that stop at step 4 of 9, and retrieval logs show the top hit ends mid-procedure. What is the most targeted fix?